A Smart Image Enhancement Method for Liver Ultrasound Elastography

Abstract

This application relates to the field of image enhancement, specifically to an intelligent image enhancement method for liver ultrasound elastography. The method includes: acquiring multiple frames of images of the patient's liver during each respiratory cycle; for each frame, determining the similarity between the frame and other frames based on the relative difference and deviation of the contour gradient of the liver region between the frame and adjacent respiratory cycles; clustering the liver region of each frame to obtain multiple clusters, and obtaining a sharpness weight for each cluster based on the gradient amplitude and motion velocity parameters; obtaining a respiratory influence coefficient based on the difference in hardening coefficient between each cluster and the corresponding cluster in similar images, the similarity between the frame and the corresponding similar images, and the mean of the sharpness weights; and enhancing the frame based on the respiratory influence coefficients of the multiple clusters in each frame to obtain the enhanced image. This application can improve the image sharpness of lesion areas.

Description

Technical Field

[0001] This application relates to the field of image enhancement technology, and in particular to an intelligent image enhancement method for liver ultrasound elastography. Background Technology

[0002] With the rapid development of non-invasive diagnostics, ultrasound elastography, which measures liver stiffness to reflect the degree of fibrosis, offers advantages such as real-time monitoring, repeatability, and no radiation. Through shear wave ultrasound elastography monitoring, the elastic and structural information of the patient's liver is fused. Based on the structural information of the patient's liver, the propagation speed of the ultrasound signal in different regions of the liver within the elastic information reflects the stiffness of different areas of the liver, i.e., the degree of fibrosis, thus achieving non-invasive detection of the patient's liver.

[0003] However, because it is necessary to fuse the elasticity information and structural information of the patient's liver, changes in thoracic air pressure during the patient's breathing movements cause deformation in some areas of the liver, which in turn affects the elasticity information of the liver, causing motion artifacts. As a result, the ultrasound elastography results after fusing the patient's elasticity information and structural information have low accuracy in some areas, making it difficult to reflect the patient's liver condition, and may even lead to misdiagnosis of the patient's condition by the attending physician. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide an intelligent image enhancement method for liver ultrasound elastography, the specific technical solution of which is as follows:

[0005] In a first aspect, a smart image enhancement method for liver ultrasound elastography is provided, the method comprising:

[0006] Multiple frames of images of the patient's liver were acquired during each respiratory cycle;

[0007] For each frame of an image in each respiratory cycle, the similarity between that frame and each frame is determined based on the relative difference and contour deviation of the contour gradient of the liver region between that frame and each frame in adjacent respiratory cycles; the liver region of each frame is obtained through semantic segmentation.

[0008] The liver region of each frame image is clustered to obtain multiple clusters. The sharpness weight of the cluster in the frame image is obtained based on the gradient magnitude in each cluster and the motion velocity parameter of the frame image. The gradient magnitude of each cluster is obtained by performing gradient filtering in the horizontal and vertical directions on the liver region to which the cluster belongs. The motion velocity parameter is obtained based on the displacement change of the preset reference point between consecutive frames.

[0009] The respiratory influence coefficient of a cluster in a given frame is obtained based on the difference in hardening coefficients between each cluster in the frame and the corresponding cluster in the similar image, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image. The hardening coefficient of each cluster is obtained based on the difference between the brightness value of each pixel in the cluster and the average brightness value of the cluster. The hardening coefficient characterizes the uniformity of hardness within the cluster. The similar images corresponding to each frame are selected from multiple frames in adjacent respiratory cycles based on similarity.

[0010] Based on the breathing influence coefficient of the multi-cluster image in each frame, image enhancement is performed on the frame to obtain the enhanced image.

[0011] Optionally, for each frame of an image in each respiratory cycle, the similarity between the frame and each other is determined based on the difference in contour gradient and contour deviation of the liver region between the frame and each frame in adjacent respiratory cycles, including:

[0012] The contour gradient of the image frame is obtained by summing the gradient magnitudes of multiple pixels on the contour of the liver region in each frame.

[0013] For each frame of an image in each respiratory cycle, the ratio of the contour gradient of that frame to the contour gradient of each frame in the adjacent respiratory cycle is calculated to obtain the contour gradient ratio of the liver region of that frame to each frame in the adjacent respiratory cycle; the relative difference of the contour gradient of the liver region between that frame and each frame in the adjacent respiratory cycle includes the contour gradient ratio.

[0014] The similarity between a given frame and other frames is determined by the ratio of the contour gradient of the liver region to that of each frame in the adjacent respiratory cycle and the contour deviation.

[0015] Optionally, the similarity between the current frame and other frames is determined based on the ratio of the contour gradient of the liver region to that of each frame in adjacent respiratory cycles and the contour deviation, including:

[0016] Calculate the Hu invariant moment distance between each frame of the image and each frame of the image in the adjacent respiratory cycle to obtain the contour deviation of the liver region between the frame of the image and each frame of the image in the adjacent respiratory cycle.

[0017] The similarity between a given frame and all other frames is determined by the ratio of the contour gradient of the liver region between each frame and the frames in adjacent respiratory cycles, and by the reciprocal of the sum of the contour deviation and a preset value.

[0018] Optionally, the liver region of each frame image is clustered to obtain multiple clusters. Based on the gradient magnitude in each cluster and the motion velocity parameters of the frame image, the sharpness weight of that cluster in the frame image is obtained, including:

[0019] The liver region of each frame of the image is clustered using the k-means clustering algorithm to obtain multiple clusters;

[0020] Gradient filtering is performed on the liver region of each frame image in both horizontal and vertical directions to obtain the horizontal gradient component and the vertical gradient component of that frame image.

[0021] For the liver region of each frame image, the gradient values ​​of multiple pixels in each cluster are obtained in the horizontal and vertical gradient components to obtain the gradient vector set of that cluster.

[0022] Calculate the mean gradient magnitude of each pixel in the gradient vector set of each cluster to obtain the gradient magnitude of that cluster;

[0023] The weight factor of each cluster is obtained by mapping the square of the gradient magnitude of each cluster through a Gaussian decay function.

[0024] The sharpness weight of a cluster in a given frame is obtained based on the motion speed parameters of each frame and the weight factor of each cluster in that frame.

[0025] Optionally, based on the motion velocity parameters of each frame and the weight factor of each cluster in that frame, the sharpness weight of that cluster in the frame is obtained, including:

[0026] For each frame of image, obtain the preset reference point on the same anatomical structure in the liver region of that frame of image and the previous frame of image;

[0027] Calculate the Euclidean distance between the preset reference point and its corresponding position in the current frame image and the previous frame image to obtain the displacement of the current frame image;

[0028] The motion speed parameters of a frame are obtained based on the displacement of each frame and the time interval between the frame and the previous frame.

[0029] The sharpness weight of a cluster in a given frame is obtained by taking the reciprocal of the sum of the motion speed parameters of each frame and a preset value, and the inverse of the weight factor of each cluster in that frame.

[0030] Optionally, before obtaining the breathing influence coefficient of a cluster in a given frame image based on the difference in hardening coefficients between each frame image and the corresponding similar image of the same cluster, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image, the method further includes:

[0031] Calculate the absolute difference between the brightness value of each pixel in each cluster and the average brightness value of that cluster, and calculate the ratio of this absolute difference to the average brightness value of that cluster to obtain the deviation value of that pixel;

[0032] Calculate the sum of the deviation values ​​of multiple pixels in each cluster to obtain the deviation value of that cluster.

[0033] Calculate the ratio of the deviation value and the sum of values ​​for each cluster to the number of pixels in that cluster, and determine the hardening coefficient for that cluster.

[0034] Optionally, before obtaining the breathing influence coefficient of a cluster in a given frame image based on the difference in hardening coefficients between each frame image and the corresponding similar image of the same cluster, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image, the method further includes:

[0035] For each frame of an image in each respiratory cycle, the similarity between that frame and each frame in adjacent respiratory cycles is calculated, and the image with the highest similarity in adjacent respiratory cycles is determined as the similar image corresponding to that frame.

[0036] Optionally, the breathing influence coefficient of a cluster in a given frame is obtained based on the difference in hardening coefficients between each cluster in each frame and the corresponding cluster in the corresponding similar image, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image, including:

[0037] Image registration is performed on each frame image and its corresponding similar images to obtain the corresponding clusters of each cluster in the frame image in the corresponding similar images;

[0038] Calculate the absolute difference of the hardening coefficient between each cluster in each frame and the corresponding cluster in the corresponding similar image to obtain the hardening difference of that cluster in that frame.

[0039] Calculate the absolute difference between the respiratory phase signal values ​​of each frame and the corresponding similar image to obtain the respiratory phase difference value of that frame.

[0040] The deformation influence coefficient of that cluster in each frame is obtained by calculating the ratio of the hardening difference of each cluster to the respiratory phase difference of that frame; the deformation influence coefficient indicates the degree of influence of respiratory motion on liver deformation.

[0041] The breathing influence coefficient of a cluster in a given frame is obtained by taking the deformation influence coefficient of each cluster in each frame, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image. The breathing influence coefficient indicates the degree of interference of breathing motion on the measurement results of the cluster.

[0042] Optionally, the breathing influence coefficient of a cluster in a given frame is obtained based on the deformation influence coefficient of each cluster in each frame, the similarity between the frame and the corresponding similar images, and the mean sharpness weight of the cluster in the frame and the corresponding similar images, including:

[0043] The initial breathing influence coefficient is obtained by multiplying the deformation influence coefficient of each cluster of each frame image, the reciprocal of the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster of the frame image and the corresponding similar image.

[0044] The initial respiratory influence coefficient is normalized to obtain the respiratory influence coefficient.

[0045] Optionally, the step of enhancing the image of each frame based on the breathing influence coefficient of multiple clusters in each frame to obtain the enhanced image includes:

[0046] Calculate the difference between the preset value and the respiratory influence coefficient of each cluster to obtain the influence weight of that cluster;

[0047] A weighted covariance matrix is ​​constructed based on the influence weights of multiple clusters in each frame of the image to perform principal component analysis on the frame of the image, thereby obtaining the enhanced elastic image.

[0048] Based on common knowledge in the field, the above-mentioned preferred conditions can be combined arbitrarily to obtain various preferred embodiments of this application.

[0049] This application offers the following advantages: Multiple ultrasound images are acquired during each respiratory cycle, and the liver region is semantically segmented in each frame. By comparing the contour gradient features and shape similarity of the liver region in adjacent respiratory cycles, corresponding similar images are selected for each frame to identify the respiratory pattern. Each liver region in each frame is clustered to obtain multiple clusters, and their displacement is tracked using a preset reference point as a stable anatomical marker to obtain motion velocity parameters. The sharpness weight is calculated by combining the gradient amplitude of each cluster, achieving accurate region registration. Based on this, a multi-parameter fusion respiratory influence parameter is constructed based on the difference in hardening coefficients of corresponding sub-regions between similar images, the overall image similarity, and the mean of the sharpness weights. The hardening coefficient reflects the uniformity of tissue hardness through the difference between the pixel brightness and the mean within a cluster. Finally, image enhancement is performed based on the respiratory influence coefficients of each cluster, suppressing areas severely affected by respiratory artifacts during enhancement, while preserving and strengthening stable, clear, and uniform tissue regions. This effectively improves the sharpness of lesion boundaries and the realism of internal hardness distribution in the reconstructed elastic image, significantly improving the image sharpness and recognition accuracy of the lesion region. Attached Figure Description

[0050] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1This is a flowchart of an intelligent image enhancement method for liver ultrasound elastography in one embodiment;

[0052] Figure 2 This is a schematic diagram of the structure of an intelligent image enhancement system for liver ultrasound elastography in one embodiment;

[0053] Figure 3 This is a schematic diagram of the structure of an electronic device in one embodiment. Detailed Implementation

[0054] To further illustrate the technical means and effects adopted by this application to achieve the intended inventive purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent image enhancement method for liver ultrasound elastography proposed according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0056] The following description, in conjunction with the accompanying drawings, details a specific scheme for an intelligent image enhancement method for liver ultrasound elastography provided in this application. For example... Figure 1 As shown, the method includes:

[0057] S11. Acquire multiple frames of images of the patient's liver during each respiratory cycle.

[0058] This application uses brightness differences to match the rate of sound wave attenuation: bright areas represent fast amplitude attenuation (loose tissue), and dark areas represent slow attenuation (dense tissue). Each scan forms an image frame, and the sound propagation speed within the area varies. At this point, the faster the transmission speed, the higher the brightness (pixel value) reflected in the area.

[0059] During each respiratory cycle, ultrasound examination was performed on the patient, acquiring multiple frames of images of the liver. Semantic segmentation was performed on each frame to identify the liver region. Simultaneously, during the acquisition of each frame, a respiratory monitoring device was used to monitor the patient's phase respiration, capturing changes in respiration during the ultrasound examination, denoted as R. The phase signal output from the respiratory monitoring device was normalized. .

[0060] During exhalation, the increased intrathoracic pressure pushes the central part of the diaphragm downwards, flattening it, while the diaphragm as a whole passively moves upwards and retracts. This series of deformations causes the liver, which is closely attached to the diaphragm, to be subjected to all-around compression, increasing internal stress and thus manifesting as increased "apparent stiffness" in elastography. Therefore, to investigate the impact of respiratory intensity during different respiratory processes on liver ultrasound examination results during ultrasound elastography, a respiratory influence coefficient was constructed and incorporated as a feature weight into the PCA (Principal Component Analysis) algorithm, thereby improving the accuracy of liver ultrasound elastography results.

[0061] S12. For each frame of an image in each respiratory cycle, determine the similarity between the frame and each other based on the relative difference and contour deviation of the contour gradient of the liver region between the frame and each frame in adjacent respiratory cycles.

[0062] In this process, the liver region of each frame is obtained through semantic segmentation, and the contour of the segmented liver region is subjected to morphological closing operation or smoothing.

[0063] During shear wave elastography, changes in intrathoracic pressure caused by the patient's respiratory movements are transmitted to the liver surface via diaphragmatic displacement, resulting in overall, low-order deformation of the liver. This type of deformation is mainly manifested as periodic changes in the liver in accordance with the respiratory rhythm and weak stress changes in the internal tissues.

[0064] During the testing process, patients may exhibit irregular breathing rhythms, resulting in the same respiratory phase not strictly corresponding to the same frame number in the time series. That is, the patient may be in the nth respiratory cycle at the same frame number. The respiratory phase corresponding to the frame image is the (n+1)th respiratory cycle. Therefore, when matching elastic maps of similar respiratory phases in patients within adjacent respiratory cycles, it is necessary to match each frame of each respiratory motion.

[0065] In one embodiment, for each frame of an image in each respiratory cycle, the similarity between the frame and each other is determined based on the difference in contour gradient and contour deviation of the liver region between the frame and each frame in adjacent respiratory cycles, including:

[0066] The contour gradient of the image frame is obtained by summing the gradient magnitudes of multiple pixels on the contour of the liver region in each frame.

[0067] For each frame of an image in each respiratory cycle, the ratio of the contour gradient of that frame to the contour gradient of each frame in the adjacent respiratory cycle is calculated to obtain the contour gradient ratio of the liver region of that frame to each frame in the adjacent respiratory cycle; the relative difference of the contour gradient of the liver region between that frame and each frame in the adjacent respiratory cycle includes the contour gradient ratio.

[0068] The similarity between a given frame and other frames is determined by the ratio of the contour gradient of the liver region to that of each frame in the adjacent respiratory cycle and the contour deviation.

[0069] Specifically, the similarity between each frame and the previous frames is determined based on the ratio of the contour gradient of the liver region to that of each frame in adjacent respiratory cycles and the contour deviation, including:

[0070] Calculate the Hu invariant moment distance between each frame of the image and each frame of the image in the adjacent respiratory cycle to obtain the contour deviation of the liver region between the frame of the image and each frame of the image in the adjacent respiratory cycle.

[0071] The similarity between a given frame and all other frames is determined by the ratio of the contour gradient of the liver region between each frame and the frames in adjacent respiratory cycles, and by the reciprocal of the sum of the contour deviation and a preset value.

[0072] The preset value can be set according to the actual situation, for example, 1.

[0073] During a single respiratory cycle, the liver undergoes slight deformation due to respiration, with its external contour changes being more pronounced than its internal changes. Specifically, the shape and circumference of the liver's external contour differ across respiratory phases. Therefore, when analyzing liver regions acquired through semantic recognition, the overall liver region similarity can be calculated based on the gradient parameters and contour length of the region edges, combined with Hu invariant moments (selecting the first seven invariant moment components). This calculation can be performed on images acquired during ultrasound examinations in adjacent respiratory cycles. Matching. The adjacent respiratory cycles of each frame indicate the next respiratory cycle of the respiratory cycle to which the current frame belongs.

[0074] No. Frame image and the first of adjacent respiratory cycles Frame image similarity The calculation formula is:

[0075] ;

[0076] in, For the first Frame image and the first of adjacent respiratory cycles Frame image similarity, For the j-th pixel, Instead of independently traversing all contour points in each of the two frames, it establishes matching point pairs based on image registration or contour point correspondence. That is: the first... The first frame outline The point, and the first The first frame outline These points are at the same location in terms of anatomical structure (e.g., obtained through equal arc length sampling, nearest neighbor mapping, or optical flow tracing). For the first The number of pixels in a frame image. The first of the adjacent respiratory cycles The number of pixels in a frame image. Indicates the first The first frame of the liver outline in the image The gradient vector at each pixel is composed of the gradient values ​​of that point in the x-direction (horizontal) and y-direction (vertical) of the image, respectively. It is usually calculated using the Sobel, Prewitt, or Scharr operators. Indicates the first of adjacent respiratory cycles The first frame of the liver outline in the image The gradient vector at each pixel is composed of the gradient values ​​of that point in the x-direction (horizontal) and y-direction (vertical) of the image. For the first The first frame of the liver region outline in the image The sum of the gradient magnitudes of the nth pixels, that is, the nth... Contour gradient of a frame image The first of the adjacent respiratory cycles The first frame of the liver region outline in the image The sum of the gradient magnitudes of the nth pixels, that is, the nth Contour gradient of a frame image It is an exponential operation with the natural constant e (approximately 2.71828) as the base.

[0077] The ratio of contour gradients in adjacent frames. The closer it is to 1, the more it indicates a frame. , The more similar the contour gradient features between them, the more they are processed by -1, and then... The inverse proportional function normalizes the result; the larger the normalized value, the larger the frame. , The higher the similarity of the outline of the mid-liver region.

[0078] For the first Frame image and the first of adjacent respiratory cycles The Hu invariant moment distance of the frame image, that is, the th frame. Frame image and the first of adjacent respiratory cycles The contour deviation of the frame image, specifically, is extracted from the first frame. Frame image and the first The first seven Hu invariant moments of the liver region contour in the frame image are used to form a feature vector with translation, rotation and scale invariance. The Euclidean distance between two feature vectors is calculated as a quantitative index of shape difference to obtain the contour deviation degree. The larger the number, the higher the number of... Frame image and the first of adjacent respiratory cycles The higher the similarity between the contours of the frame images, the more "1+" is used to avoid the result change being in the opposite direction to the similarity change. To pass the first Frame image and the first The similarity of liver region contours in frame images is analyzed to obtain the similarity of liver deformation caused by the same respiratory phase in different respiratory movements, thereby improving the accuracy of liver internal analysis in different respiratory phases.

[0079] S13. Cluster the liver region of each frame image to obtain multiple clusters. Based on the gradient magnitude in each cluster and the motion speed parameter of the frame image, obtain the sharpness weight of the cluster of the frame image.

[0080] The gradient magnitude of each cluster is obtained by performing gradient filtering in the horizontal and vertical directions on the liver region to which the cluster belongs, and the motion velocity parameter is obtained based on the displacement change of the preset reference point between consecutive frames.

[0081] The liver is the largest and highest-located solid organ in the abdominal cavity. When subjected to external pressure, such as increased thoracic pressure during exhalation causing the diaphragm to move downwards and compress the liver, the liver experiences direct physical compression, altering its internal stress state and leading to a temporary increase in measured apparent stiffness values ​​(such as Young's modulus). Conversely, during inhalation, the liver relaxes, and the apparent stiffness value may decrease. However, due to factors such as the patient's physical condition or anxiety during the examination, the respiratory rate may become disordered and irregular, resulting in poor removal of motion artifacts by the anti-motion artifact algorithm and affecting the clarity of the elastogram.

[0082] During the shear wave elastography detection of patients, it is necessary to analyze the patient's respiratory rate, identify the detection period when the patient's respiratory rate is disordered, and analyze the regional characteristics of each frame of elastography within that period.

[0083] The location of the hepatic artery in the liver geometry of each patient is similar in macroscopic functional framework. However, the deformation of the liver caused by the patient's respiratory movements is mainly concentrated in the area near the diaphragm. When the deformation is transmitted to the inside of the liver, it is severely weakened. At this time, the impact on the inside of the liver is relatively small. Therefore, based on anatomical prior knowledge, relatively stable hyperechoic tubular structures inside the liver (such as the portal vein or the main trunk of the hepatic artery) are identified as preset reference points. The optical flow field is weighted and corrected using their relative displacement. If the preset reference point is not detected in the current frame, the average motion vector of optical flow in the whole image or the liver region is calculated as a substitute. The optical flow field is corrected based on the location of the hepatic artery, and the regional coordinate transformation is used to achieve matching of different liver regions in connected frames.

[0084] In one embodiment, the liver region of each frame image is clustered to obtain multiple clusters. Based on the gradient magnitude in each cluster and the motion velocity parameters of the frame image, the sharpness weight of that cluster in the frame image is obtained, including:

[0085] The liver region of each frame of the image is clustered using the k-means clustering algorithm to obtain multiple clusters;

[0086] Gradient filtering is performed on the liver region of each frame image in both horizontal and vertical directions to obtain the horizontal gradient component and the vertical gradient component of that frame image.

[0087] For the liver region of each frame image, the gradient values ​​of multiple pixels in each cluster are obtained in the horizontal and vertical gradient components to obtain the gradient vector set of that cluster.

[0088] Calculate the mean gradient magnitude of each pixel in the gradient vector set of each cluster to obtain the gradient magnitude of that cluster;

[0089] The weight factor of each cluster is obtained by mapping the square of the gradient magnitude of each cluster through a Gaussian decay function.

[0090] The sharpness weight of a cluster in a given frame is obtained based on the motion speed parameters of each frame and the weight factor of each cluster in that frame.

[0091] Specifically, based on the motion velocity parameters of each frame and the weight factor of each cluster in that frame, the sharpness weight of that cluster in the frame is obtained, including:

[0092] For each frame of image, obtain the preset reference point on the same anatomical structure in the liver region of that frame of image and the previous frame of image;

[0093] Calculate the Euclidean distance between the preset reference point and its corresponding position in the current frame image and the previous frame image to obtain the displacement of the current frame image;

[0094] The motion speed parameters of a frame are obtained based on the displacement of each frame and the time interval between the frame and the previous frame.

[0095] The sharpness weight of a cluster in a given frame is obtained by taking the reciprocal of the sum of the motion speed parameters of each frame and a preset value, and the inverse of the weight factor of each cluster in that frame.

[0096] During ultrasound imaging of a patient, each frame of the image is captured. When the patient exhales, the diaphragm moves due to changes in thoracic cavity pressure, causing the liver to be compressed and undergo elastic deformation. At this time, during shear wave ultrasound monitoring, there are differences in brightness (elastic modulus value E of each pixel) between different regions. Therefore, based on the brightness differences between different pixels in the liver region, the k-means clustering algorithm is used to perform mean clustering to obtain multiple clusters, each denoted as k.

[0097] To analyze similar frame pairs and identify areas with less artifacts caused by respiratory motion in the same respiratory phase, it is necessary to compare the differences in boundary sharpness between different regions (clusters) within the liver. Higher sharpness indicates a lower severity of artifacts caused by respiratory motion. Therefore, a local sharpness weight is applied to each clustered region within the liver. The model is constructed and used as a reference for region filtering in similar frame pairs.

[0098] Gradient filtering operations in the horizontal and vertical directions are applied to the segmented liver region in each frame of the image to obtain the horizontal gradient component map and the vertical gradient component map corresponding to the liver region. Then, for each cluster divided by the clustering algorithm, the gradient values ​​of all pixels in the cluster corresponding to the horizontal and vertical gradient component maps are extracted, thereby forming the gradient vector set of the cluster. Each vector in the gradient vector set contains the gradient value of the pixel in the horizontal and vertical directions. Finally, the arithmetic mean of the gradient magnitudes of all pixels in the gradient vector set is calculated, and this mean is determined as the gradient magnitude of the cluster.

[0099] For each frame of the ultrasound image sequence (except the first frame), a pre-defined reference point on the same stable anatomical structure within the liver region of that frame and the previous frame is identified and located. The pre-defined reference point is preferably a easily identifiable feature location within the main trunk of the hepatic artery or portal vein that exhibits high displacement consistency. Next, the Euclidean distance between the corresponding coordinates of the pre-defined reference point in the two frames is calculated, serving as the inter-frame displacement reflecting the overall movement of the liver due to respiration. Finally, the ratio of the displacement to the actual time interval between the acquisition of the two frames is calculated to obtain the motion velocity parameter of that frame. This parameter quantifies the instantaneous rate of respiratory motion within the imaging plane and is used to characterize the risk of blurring due to global motion in subsequent sharpness weighting calculations.

[0100] No. The first frame of the image Clarity weight of each cluster The calculation formula is:

[0101] ;

[0102] in, For the first The first frame of the image The clarity weight of each cluster, For the first The first frame of the image The gradient magnitude of each cluster, For the first The first frame of the image The sum of squares of the horizontal and vertical gradient components of each cluster. The larger the value, the sharper the image edges and the clearer the details. (represents the square of the L2 norm of a vector). The standard deviation and variance of the Gaussian distribution can be set empirically, for example, by taking a value that is 0.5 to 2 times the mean of the gradient of the entire graph. The weighting factor is obtained by mapping using a Gaussian decay function. The local sharpness coefficient is obtained by comparing the local sharpness with a sharpness threshold. For the first Motion velocity parameters of the frame image For the first The absolute value of the motion speed parameter of the frame image; the faster the motion, the higher the velocity. The lower the value, the less likely it is to cause ghosting frames to participate in the reconstruction. To find the minimum value, avoid a denominator of 0, for example... for or , The analysis yields locally clear parameters of the patient's respiratory motion within the cluster; the higher the value, the less the current cluster is affected by respiratory motion.

[0103] S14. Based on the difference in hardening coefficients between each cluster in each frame image and the corresponding cluster in the corresponding similar image, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image, the breathing influence coefficient of the cluster in the frame image is obtained.

[0104] The hardening coefficient of each cluster is obtained based on the difference between the brightness value of each pixel in the cluster and the average brightness value of the cluster. The hardening coefficient characterizes the uniformity of hardness within the cluster.

[0105] Similar images corresponding to each frame are selected from multiple frames of adjacent respiratory cycles based on their similarity.

[0106] When analyzing the impact of respiratory movements on the liver, in addition to analyzing the main affected locations such as the regional boundaries, the overall performance within the region is also an important influencing factor. As the main object of analysis for patient symptoms, the brightness of each region within the liver reflects the degree of liver sclerosis. Therefore, based on the brightness values ​​within each cluster, local regional sclerosis characteristics are constructed according to the brightness differences and the size of the cluster region.

[0107] Therefore, in one embodiment, before obtaining the breathing effect coefficient of the cluster of the frame image based on the difference in hardening coefficients between each frame image and the corresponding similar image of the same cluster, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster of the frame image and the corresponding similar image, the method further includes:

[0108] Calculate the absolute difference between the brightness value of each pixel in each cluster and the average brightness value of that cluster, and calculate the ratio of this absolute difference to the average brightness value of that cluster to obtain the deviation value of that pixel;

[0109] Calculate the sum of the deviation values ​​of multiple pixels in each cluster to obtain the deviation value of that cluster.

[0110] Calculate the ratio of the deviation value and the sum of values ​​for each cluster to the number of pixels in that cluster, and determine the hardening coefficient for that cluster.

[0111] Hardening coefficient per cluster The calculation formula is:

[0112] ;

[0113] in, For the first The hardening coefficient of each cluster, For the first Within the cluster, the first 1 pixel For the first The number of pixels in a cluster For the first The average brightness value of multiple pixels in a cluster. For the first Within the cluster, the first The brightness value (pixel value) of each pixel. This indicates the difference in brightness among pixels within a cluster. The larger the value, the greater the possibility of artifacts produced by respiratory motion within the cluster. It represents the overall feature relationship of pixels within a clustered region and serves as a quantification parameter for the clustered region features. To assign attention to large areas of abnormal stiffness, an area weight was introduced as a parameter to account for the influence of respiratory movements on regional deformation of the liver. To find the minimum value, avoid a denominator of 0, for example... for or . To measure the uniformity of hardness within a cluster and the overall importance of the region.

[0114] In one embodiment, before obtaining the breathing effect coefficient of a cluster in a frame image based on the difference in hardening coefficients between each frame image and the corresponding similar image of the same cluster, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image, the method further includes:

[0115] For each frame of an image in each respiratory cycle, the similarity between that frame and each frame in adjacent respiratory cycles is calculated, and the image with the highest similarity in adjacent respiratory cycles is determined as the similar image corresponding to that frame.

[0116] Similarity between frames of adjacent respiratory movements during ultrasound elastography of patients Perform matching, and select the nth breath cycle from the nth breath cycle. The frame image is matched with each frame of the (n+1)th respiratory cycle to obtain the... Maximum similarity of frame images And the image corresponding to the (n+1)th respiratory cycle is used as the first... Similar images corresponding to the frame image, the first A frame image and its corresponding similar image form a similar frame pair.

[0117] In one embodiment, the breathing effect coefficient of a cluster in a frame is obtained based on the difference in hardening coefficients between each cluster in each frame and the corresponding cluster in the corresponding similar image, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image, including:

[0118] Image registration is performed on each frame image and its corresponding similar images to obtain the corresponding clusters of each cluster in the frame image in the corresponding similar images;

[0119] Calculate the absolute difference of the hardening coefficient between each cluster in each frame and the corresponding cluster in the corresponding similar image to obtain the hardening difference of that cluster in that frame.

[0120] Calculate the absolute difference between the respiratory phase signal values ​​of each frame and the corresponding similar image to obtain the respiratory phase difference value of that frame.

[0121] The deformation influence coefficient of that cluster in each frame is obtained by calculating the ratio of the hardening difference of each cluster to the respiratory phase difference of that frame; the deformation influence coefficient indicates the degree of influence of respiratory motion on liver deformation.

[0122] The breathing influence coefficient of a cluster in a given frame is obtained by taking the deformation influence coefficient of each cluster in each frame, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image. The breathing influence coefficient indicates the degree of interference of breathing motion on the measurement results of the cluster.

[0123] Specifically, based on the deformation influence coefficient of each cluster in each frame image, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image, the breathing influence coefficient of the cluster in the frame image is obtained, including:

[0124] The initial breathing influence coefficient is obtained by multiplying the deformation influence coefficient of each cluster of each frame image, the reciprocal of the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster of the frame image and the corresponding similar image.

[0125] The initial respiratory influence coefficient is normalized to obtain the respiratory influence coefficient.

[0126] During respiratory movements, different regions of the liver are affected to varying degrees, and the degree of influence varies in different respiratory phases. To improve the accuracy of region recognition, the matching similarity between corresponding clusters of liver regions in connected frames is used, and parameters of liver deformation affected by respiration are obtained by combining the depth changes during the respiratory process.

[0127] Image registration is performed on each frame of the image and its corresponding similar images to obtain the corresponding cluster of each cluster in the corresponding similar images for each cluster of the frame image. Specifically, for the first cluster in the frame image... Each cluster is mapped to the coordinate system of a similar image through a registration transformation. The cluster with the highest spatial overlap or the closest geometric features is found in the clustering results of similar images and used as its corresponding cluster for subsequent cross-frame feature comparison.

[0128] Breathing effect coefficient of the k-th cluster in each frame The calculation formula is:

[0129] ;

[0130] in, The breathing effect coefficient of the k-th cluster in each frame of the image. For the first The first frame of the image The hardening coefficient of each cluster, For the first The first frame of the image The hardening coefficient of the cluster, the first The first frame of the image The cluster is the first The first frame of the image The corresponding cluster of each cluster, For the first The respiratory phase difference value of the frame image, that is, the first frame image. Frame image and the first The absolute difference between the respiratory phase signal values ​​of the frame images. To find the minimum value, avoid a denominator of 0, for example... for or , For the first The frame image and its corresponding similar image, that is, the first frame image Frame image similarity, For the first The frame image and its corresponding similar image, that is, the first frame image The average value of the sharpness weight of the frame image. This is a normalization function used to normalize values ​​to the interval [0,1].

[0131] The ratio of the hardening difference between corresponding regions of adjacent frames to the respiratory phase difference reflects the influence of a unit respiratory phase on liver organ deformation during respiratory operation. The larger the value, the greater the influence of the patient's breathing on liver deformation. For frames , The reciprocal of the similarity between them is used here as a global influencing factor for the liver, serving as an overall constraint to reduce the risk of abnormal matching in local regions. For frames , The mean of the sharpness weights between them, through Obtain the blur parameters for the current region. The higher the value, the lower the clarity of the clustered region boundaries, and the greater the likelihood that the liver will produce motion artifacts due to respiratory effects. This is the elasticity change coefficient of the clustered region under respiratory disturbance. The larger the value, the higher the possibility that the data in this region will become invalid due to respiratory disturbance.

[0132] S15. Based on the breathing influence coefficient of the multi-cluster image in each frame, perform image enhancement on the frame to obtain the enhanced image.

[0133] In one embodiment, image enhancement is performed on the frame image based on the breathing effect coefficient of the multi-cluster image to obtain an enhanced image, including:

[0134] Calculate the difference between the preset value and the respiratory influence coefficient of each cluster to obtain the influence weight of that cluster;

[0135] A weighted covariance matrix is ​​constructed based on the influence weights of multiple clusters in each frame of the image to perform principal component analysis on the frame of the image, thereby obtaining the enhanced elastic image.

[0136] Comparison of respiratory influence parameters between similar frames within different respiratory motion cycles The higher the value (the closer to 1), the more likely the cluster is to be identified as a breathing artifact; conversely, a lower value may indicate a true anatomical structure. This is used to determine the cluster's influence weight. Influence weight The calculation formula is: , This is the respiration effect coefficient of the cluster.

[0137] Weight the influence of multiple clusters in each frame of the image. We introduce weighted PCA to construct a weighted covariance matrix. ,in The total number of clusters, For the first The feature vectors of each cluster Its weighted mean, This is the vector transpose. And the weighted covariance matrix is... Feature decomposition is performed, selecting the top few principal components with the highest energy (usually the first principal component). Since the elastic distribution of real liver tissue exhibits spatial consistency, its main energy is concentrated in the first one or two principal components, while respiratory artifacts typically appear as low-energy, high-frequency perturbations. Therefore, only the top few principal components with the highest energy are selected. Principal components (usually) By reconstructing the image, the true elastic structure of the liver can be effectively preserved. Based on these principal components, the original image is reconstructed, thereby effectively suppressing respiratory artifacts and obtaining an enhanced image.

[0138] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0139] This application also provides an intelligent image enhancement system for liver ultrasound elastography, such as... Figure 2 As shown, the system includes:

[0140] Acquisition module 21 is used to acquire multiple frames of images of the patient's liver during each respiratory cycle;

[0141] The determination module 22 is used to determine the similarity between each frame of an image and each frame of an image in each respiratory cycle, based on the relative difference and contour deviation of the contour gradient of the liver region between the frame of an image and each frame of an image in adjacent respiratory cycles; the liver region of each frame of an image is obtained through semantic segmentation.

[0142] Clustering module 23 is used to cluster the liver region of each frame image to obtain multiple clusters. Based on the gradient magnitude in each cluster and the motion speed parameter of the frame image, the sharpness weight of the cluster in the frame image is obtained. The gradient magnitude of each cluster is obtained by performing gradient filtering in the horizontal and vertical directions on the liver region to which the cluster belongs. The motion speed parameter is obtained based on the displacement change of the preset reference point between consecutive frames.

[0143] The calculation module 24 is used to obtain the breathing influence coefficient of a cluster in a frame image based on the difference in hardening coefficient between each cluster in each frame image and the corresponding cluster in the corresponding similar image, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image. The hardening coefficient of each cluster is obtained based on the difference between the brightness value of each pixel in the cluster and the average brightness value of the cluster. The hardening coefficient characterizes the uniformity of hardness within the cluster. The similar images corresponding to each frame image are selected from multiple frames of adjacent respiratory cycles based on the similarity.

[0144] The enhancement module 25 is used to enhance the image of each frame based on the breathing influence coefficient of the multi-cluster image, so as to obtain the enhanced image.

[0145] For the system embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this application according to actual needs.

[0146] Figure 3 This is a schematic diagram of the structure of an electronic device according to an example embodiment of this application. The electronic device includes a memory, a processor, and a computer program stored in the memory and used to run on the processor. When the processor executes the computer program, it implements the method described in any of the above embodiments. Figure 3The electronic device 30 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0147] like Figure 3 As shown, the electronic device 30 can be manifested as a general-purpose computing device, such as a server device. The components of the electronic device 30 may include, but are not limited to: at least one processor 31, at least one memory 32, and a bus 33 connecting different system components (including memory 32 and processor 31).

[0148] Bus 33 includes a data bus, an address bus, and a control bus.

[0149] The memory 32 may include volatile memory, such as random access memory (RAM) 321 and / or cache memory 322, and may further include read-only memory (ROM) 323.

[0150] The memory 32 may also include a program tool 325 (or utility) having a set (at least one) program module 324, such program module 324 including but not limited to: an operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0151] The processor 31 executes various functional applications and data processing, such as the methods provided in any of the above embodiments, by running computer programs stored in the memory 32.

[0152] Electronic device 30 can also communicate with one or more external devices 34 (e.g., keyboard, pointing device, etc.). This communication can be performed via input / output (I / O) interface 35. Furthermore, electronic device 30 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public network, such as the Internet) via network adapter 36. As shown, network adapter 36 communicates with other modules of electronic device 30 via bus 33. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with electronic device 30, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems.

[0153] It should be noted that although several units / modules or sub-units / modules of the electronic device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.

[0154] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided in any of the above embodiments.

[0155] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0156] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0157] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the method described in any of the above embodiments.

[0158] The program code for executing the computer program product of this application can be written in any combination of one or more programming languages. The program code can be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0159] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0160] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

[0161] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. An intelligent image enhancement method for liver ultrasound elastography, characterized in that, The method includes: Multiple frames of images of the patient's liver were acquired during each respiratory cycle; For each frame of an image in each respiratory cycle, the similarity between that frame and each frame is determined based on the relative difference and contour deviation of the contour gradient of the liver region between that frame and each frame in adjacent respiratory cycles; the liver region of each frame is obtained through semantic segmentation. The liver region of each frame image is clustered to obtain multiple clusters. The sharpness weight of the cluster in the frame image is obtained based on the gradient magnitude in each cluster and the motion velocity parameter of the frame image. The gradient magnitude of each cluster is obtained by performing gradient filtering in the horizontal and vertical directions on the liver region to which the cluster belongs. The motion velocity parameter is obtained based on the displacement change of the preset reference point between consecutive frames. The respiratory influence coefficient of a cluster in a given frame is obtained based on the difference in hardening coefficients between each cluster in the frame and the corresponding cluster in the similar image, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image. The hardening coefficient of each cluster is obtained based on the difference between the brightness value of each pixel in the cluster and the average brightness value of the cluster. The hardening coefficient characterizes the uniformity of hardness within the cluster. The similar images corresponding to each frame are selected from multiple frames in adjacent respiratory cycles based on similarity. Based on the breathing influence coefficient of the multi-cluster image in each frame, image enhancement is performed on the frame to obtain the enhanced image.

2. The intelligent image enhancement method for liver ultrasound elastography of claim 1, wherein, For each frame of an image in each respiratory cycle, the similarity between that frame and the previous frames is determined based on the difference in contour gradient and contour deviation of the liver region between that frame and the previous frames in adjacent respiratory cycles, including: The contour gradient of the image frame is obtained by summing the gradient magnitudes of multiple pixels on the contour of the liver region in each frame. For each frame of an image in each respiratory cycle, the ratio of the contour gradient of that frame to the contour gradient of each frame in the adjacent respiratory cycle is calculated to obtain the contour gradient ratio of the liver region of that frame to each frame in the adjacent respiratory cycle; the relative difference of the contour gradient of the liver region between that frame and each frame in the adjacent respiratory cycle includes the contour gradient ratio. The similarity between a given frame and other frames is determined by the ratio of the contour gradient of the liver region to that of each frame in the adjacent respiratory cycle and the contour deviation.

3. The intelligent image enhancement method for liver ultrasound elastography of claim 2, wherein, The method of determining the similarity between a given frame and other frames based on the ratio of the contour gradient of the liver region to that of adjacent frames in a respiratory cycle and the contour deviation includes: Calculate the Hu invariant moment distance between each frame of the image and each frame of the image in the adjacent respiratory cycle to obtain the contour deviation of the liver region between the frame of the image and each frame of the image in the adjacent respiratory cycle. The similarity between a given frame and all other frames is determined by the ratio of the contour gradient of the liver region between each frame and the frames in adjacent respiratory cycles, and by the reciprocal of the sum of the contour deviation and a preset value.

4. The intelligent image enhancement method for liver ultrasound elastography of claim 1, wherein, The process involves clustering the liver region of each frame image to obtain multiple clusters. Based on the gradient magnitude within each cluster and the motion velocity parameters of the frame image, a sharpness weight for that cluster is determined, including: The liver region of each frame of the image is clustered using the k-means clustering algorithm to obtain multiple clusters; Gradient filtering is performed on the liver region of each frame image in both horizontal and vertical directions to obtain the horizontal gradient component and the vertical gradient component of that frame image. For the liver region of each frame image, the gradient values ​​of multiple pixels in each cluster are obtained in the horizontal and vertical gradient components to obtain the gradient vector set of that cluster. Calculate the mean gradient magnitude of each pixel in the gradient vector set of each cluster to obtain the gradient magnitude of that cluster; The weight factor of each cluster is obtained by mapping the square of the gradient magnitude of each cluster through a Gaussian decay function. The sharpness weight of a cluster in a given frame is obtained based on the motion speed parameters of each frame and the weight factor of each cluster in that frame.

5. An intelligent image enhancement method for liver ultrasound elastography as claimed in claim 4, wherein, The step of obtaining the sharpness weight of a cluster in a frame image based on the motion speed parameters of each frame image and the weight factor of each cluster in that frame image includes: For each frame of image, obtain the preset reference point on the same anatomical structure in the liver region of that frame of image and the previous frame of image; Calculate the Euclidean distance between the preset reference point and its corresponding position in the current frame image and the previous frame image to obtain the displacement of the current frame image; The motion speed parameters of a frame are obtained based on the displacement of each frame and the time interval between the frame and the previous frame. The sharpness weight of a cluster in a given frame is obtained by taking the reciprocal of the sum of the motion speed parameters of each frame and a preset value, and the inverse of the weight factor of each cluster in that frame.

6. The intelligent image enhancement method for liver ultrasound elastography of claim 1, wherein, Before obtaining the breathing effect coefficient of a cluster in a given frame image based on the difference in hardening coefficients between each frame image and the corresponding similar image of the same cluster, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image, the method further includes: Calculate the absolute difference between the brightness value of each pixel in each cluster and the average brightness value of that cluster, and calculate the ratio of this absolute difference to the average brightness value of that cluster to obtain the deviation value of that pixel; Calculate the sum of the deviation values ​​of multiple pixels in each cluster to obtain the deviation value of that cluster. Calculate the ratio of the deviation value and the sum of values ​​for each cluster to the number of pixels in that cluster, and determine the hardening coefficient for that cluster.

7. An intelligent image enhancement method for liver ultrasound elastography as claimed in claim 6, wherein, Before obtaining the breathing effect coefficient of a cluster in a given frame image based on the difference in hardening coefficients between each frame image and the corresponding similar image of the same cluster, the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster in the frame image and the corresponding similar image, the method further includes: For each frame of an image in each respiratory cycle, the similarity between that frame and each frame in adjacent respiratory cycles is calculated, and the image with the highest similarity in adjacent respiratory cycles is determined as the similar image corresponding to that frame.

8. The intelligent image enhancement method for liver ultrasound elastography of claim 7, wherein, The breathing effect coefficient of a cluster in a given frame is obtained based on the difference in hardening coefficients between each cluster in each frame and the corresponding cluster in the corresponding similar image, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image. This includes: Image registration is performed on each frame image and its corresponding similar images to obtain the corresponding clusters of each cluster in the frame image in the corresponding similar images; Calculate the absolute difference of the hardening coefficient between each cluster in each frame and the corresponding cluster in the corresponding similar image to obtain the hardening difference of that cluster in that frame. Calculate the absolute difference between the respiratory phase signal values ​​of each frame and the corresponding similar image to obtain the respiratory phase difference value of that frame. The deformation influence coefficient of that cluster in each frame is obtained by calculating the ratio of the hardening difference of each cluster to the respiratory phase difference of that frame; the deformation influence coefficient indicates the degree of influence of respiratory motion on liver deformation. The breathing influence coefficient of a cluster in a given frame is obtained by taking the deformation influence coefficient of each cluster in each frame, the similarity between the frame and the corresponding similar image, and the mean sharpness weight of the cluster in the frame and the corresponding similar image. The breathing influence coefficient indicates the degree of interference of breathing motion on the measurement results of the cluster.

9. The intelligent image enhancement method for liver ultrasound elastography of claim 8, wherein, The method of obtaining the breathing influence coefficient of a cluster in a frame image based on the deformation influence coefficient of each cluster in each frame image, the similarity between the frame image and the corresponding similar images, and the mean sharpness weight of the cluster in the frame image and the corresponding similar images includes: The initial breathing influence coefficient is obtained by multiplying the deformation influence coefficient of each cluster of each frame image, the reciprocal of the similarity between the frame image and the corresponding similar image, and the mean sharpness weight of the cluster of the frame image and the corresponding similar image. The initial respiratory influence coefficient is normalized to obtain the respiratory influence coefficient.

10. The intelligent image enhancement method for liver ultrasound elastography of claim 1, wherein, The step of enhancing the image frame based on the breathing influence coefficient of the multi-cluster image in each frame to obtain the enhanced image includes: Calculate the difference between the preset value and the respiratory influence coefficient of each cluster to obtain the influence weight of that cluster; A weighted covariance matrix is ​​constructed based on the influence weights of multiple clusters in each frame of the image to perform principal component analysis on the frame of the image, thereby obtaining the enhanced elastic image.

Images